Analysis of a Bluetooth Low Energy - Indoor Localization System (BLE-ILS) on BLE 5.0 and Bluetooth 3.0

This study focused on developing a Bluetooth Low Energy indoor localization system (BLE-ILS) to track objects in indoor applications. Using fingerprinting with Random Forests (RF) and k-nearest Neighbors (k-NN), the system's performance was evaluated in terms of error metrics and resource consumption. Moreover, the models were tested on mobile devices with two different Bluetooth versions, Bluetooth 3.0 and BLE 5.0. The system development includes the use of XIAO BLE Sense as anchors, Bluetooth 3.0 and BLE 5.0 phones as beacons, and an Intel Gen10 i5 processor with 16GB RAM laptop running in Windows 10 that served as the gateway and compute server. Results have shown better accuracy for RF compared to kNN. Additionally, a model trained solely in BLE 5.0 had a significant increase in error when tested with Bluetooth 3.0. Furthermore, a retrained model enhanced the accuracy from 3.35m to 2.5m for Bluetooth 3.0 data even with reduced sample sizes compared to the initial model. In addition, the number of anchors can be reduced without significantly affecting performance with the correct anchor layout; however, it was also observed that the ILS's performance was significantly affected by the number of simultaneously operating beacons.